{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "b6948f5d-bf4b-4704-8249-0bfe965bcccc",
   "metadata": {},
   "outputs": [],
   "source": [
    "def return_eval(pred2score, target2score, mean):\n",
    "    mean2 = 2 * mean\n",
    "    pred  = [p.lower() for p in pred2score]\n",
    "    target = [p.lower() for p in target2score]\n",
    "    o = len(set(target))\n",
    "\n",
    "    intersect = len(set(pred[:o]).intersection(set(target)))\n",
    "    budgetaccone = len(set(pred[:mean]).intersection(set(target)))/mean\n",
    "    budgetacctwo = len(set(pred[:mean2]).intersection(set(target)))/mean2\n",
    "    prec = intersect/len(set(pred[:o])) if len(pred) > 0 else 0.0\n",
    "    rec = intersect/len(target)\n",
    "\n",
    "    \n",
    "    kmean = len(set(pred[:mean]))\n",
    "    k2mean = len(set(pred[:mean2]))\n",
    "\n",
    "    if prec==0 and rec==0:\n",
    "        f1=0\n",
    "    else:\n",
    "        f1 = 2*prec*rec/(prec+rec)\n",
    "    \n",
    "    return {\"P@O\":100*prec, \"R@O\": 100*rec, \"F1@O\":100*f1, \"B@mean\": budgetaccone, \"B@2mean\": budgetacctwo, \"#k@mean\": kmean, \"#k@2mean\": k2mean}\n",
    "\n",
    "def final_metric_results(preds_keyphrases, labels_keyphrases, mean):\n",
    "    avg_scores = defaultdict(list)\n",
    "    for pred, target in zip(preds_keyphrases, labels_keyphrases):\n",
    "\n",
    "            all_exact_results = return_eval(pred, target, mean)\n",
    "            \n",
    "            for m_name, value in all_exact_results.items():\n",
    "                    avg_scores[m_name].append(value)\n",
    "\n",
    "            avg_scores[\"pred_kpnum\"].append(len(set(pred)))\n",
    "            avg_scores[\"gt_kpnum\"].append(len(set(target)))\n",
    "            \n",
    "    avg_scores = {m_name: round(np.mean(values),2) for m_name, values in avg_scores.items()}\n",
    "\n",
    "    return avg_scores\n",
    "    \n",
    "def generate_results(df, mean):\n",
    "    \n",
    "    labels_keyphrases = [p.lower().split(\";\") for p in df[\"target\"]]\n",
    "    preds_keyphrases = []\n",
    "    for i in range(len(df)):\n",
    "        # preds_keyphrases.append(post_process(df.iloc[i][\"keyword\"])[:k])\n",
    "        preds_keyphrases.append(post_process(df.iloc[i][\"keyword\"]))\n",
    "        \n",
    "    print(\"@\",mean) \n",
    "    return final_metric_results(preds_keyphrases, labels_keyphrases, mean)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "761906b8-6c1d-4d48-adcf-20589d9a0385",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}